非线性系统的梯度变分迭代自学习控制

研究了一类非线性系统的梯度变分迭代自学习算法，以提高此类非线性系统的控制品质.梯度变分迭代自学习算法是针对符合某一类范式的周期性或重复性输出控制的非线性系统而设计的一种自寻优自学习算法.该算法针对一类非线性系统的数学描述模型，给出了性能指标函数，通过梯度变分的方法寻找性能指标函数梯度的负方向，并利用迭代自学习得到性能指标函数的最小值，使系统收敛于目标输出.将该算法应用于极端环境模拟装置的压力控制系统，取得了比传统控制算法更高的效率与更快的收敛速度.梯度变分迭代自学习算法是符合一类数学模型的非线性系统的一种高效控制算法.

Gradient-variation iterative learning control for nonlinear systems

A gradient-variation iteration learning control algorithm for one kind of nonlinear systems was discussed to improve the control quality.This algorithm was designed to be used in periodic and repeated output nonlinear systems that conform to a certain paradigm.The algorithm gives the function of the performance index for the mathematical model of the nonlinear system.The algorithm can find out the negative direction of grads of the performance index function by the gradient-variation method,and get the minimum of the performance index function by the iteration learning method,which makes the system converge to the target output.This algorithm was applied successfully to the pressure control system of an extreme stress environmental analog device with higher efficiency and faster convergence speed than those of the traditional control algorithm.The test results indicate that the gradient-variation iteration learning control algorithm is an effective control algorithm for nonlinear systems that conform to a certain mathematical model.